---
name: notion-to-blog
description: Transfer a blog post from Notion to the Wasp blog. Fetches content, downloads and optimizes images, and creates a properly formatted MDX file.
argument-hint: "[notion-page-url]"
---

## Transfer a Notion page to the Wasp blog

Given a Notion page URL, convert it into an MDX blog post for the Wasp website at `web/blog/`.

### Step 1: Fetch the Notion page

1. Extract the page ID from the URL (the UUID at the end, formatted with dashes: `xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx`).
2. Use `mcp__notion__API-retrieve-a-page` to get page metadata (title, properties).
3. Use `mcp__notion__API-get-block-children` to fetch all content blocks. Notion paginates at 100 blocks — check `has_more` and `next_cursor` to fetch all pages.

### Step 2: Parse the content

The Notion page may contain WIP/working notes before and after the actual blog post. Ask the user where the actual post starts and ends if unclear. Typically:
- The post starts at the first `heading_1` block
- WIP sections are often marked with a `## WIP` heading or appear after a `divider` near the end

Use a Python script (via Bash) to parse the JSON block data because it can be very large (50K+ tokens). The script should:
- Extract blocks in the post range
- Convert each block to markdown based on its type:
  - `heading_1/2/3` → `# / ## / ###`
  - `paragraph` → plain text with formatting
  - `bulleted_list_item` → `- `
  - `numbered_list_item` → `1. `
  - `code` → fenced code block with language
  - `image` → collect URL and caption
  - `quote` → `> `
  - `callout` → `> {emoji} `
  - `divider` → `---`
  - `table_of_contents` → skip (Docusaurus handles this)
- Preserve rich text formatting: bold (`**`), italic (`*`), code (`` ` ``), strikethrough (`~~`), links (`[text](url)`)
- Collect all image URLs and captions separately

### Step 3: Download and optimize images

1. Create directory: `web/static/img/<post-slug>/`
2. Download all images using `curl` (Notion URLs are temporary signed S3 URLs — download immediately).
3. Give images descriptive names based on their context/caption.
4. Convert all images to WebP using `cwebp -q 85` for smaller file sizes.
5. Remove the original files after conversion.

### Step 4: Create the MDX file

Create `web/blog/YYYY-MM-DD-<slug>.mdx` following these conventions:

**Frontmatter:**
```yaml
---
title: "Post Title"
authors: [authorHandle]  # from web/blog/authors.yml
image: /img/<post-slug>/banner.webp
tags: [wasp, tag2, tag3]  # 3-6 lowercase tags
keywords: ["keyword1", "keyword2", ...]  # 10+ for SEO
description: "SEO summary under 160 chars."
---
```

**Content rules:**
- Add `import { ImgWithCaption } from './components/ImgWithCaption';` after frontmatter
- Add `{/* truncate */}` after the opening hook paragraph (this controls the blog listing excerpt)
- Use `<ImgWithCaption source="..." caption="..." alt="..." />` for ALL images. Omit `caption` prop if there's no caption. When a caption contains double quotes, use single quotes for the prop value instead: `caption='text with "quotes" in it'`
- Make ALL URLs pointing to wasp.sh or wasp-lang.dev **relative** (e.g., `https://wasp.sh/blog/...` → `/blog/...`, `https://wasp.sh/docs/...` → `/docs/...`, `https://wasp-lang.dev/docs/...` → `/docs/...`). This applies to both markdown links and any other references throughout the entire post.
- Keep external URLs (GitHub, Twitter, etc.) as-is

### Step 5: Verify with the user

Show a summary of what was created:
- File path
- Number of images downloaded and total size savings from WebP conversion
- Any decisions made (title choice, skipped sections, etc.)
- Remind user to review anchor links in TL;DR sections